Sampling techniques for Arabic Sentiment Classification: A Comparative Study

Over the last decade, the web 2.0 has been shifting the web to turn it into an opinion platform. This results in a large amount of raw data that overwhelms human capacity to extract valuable knowledge without assistance of machines. In real world applications, sentiment analysis faces imbalanced data problem. To tackle this problem, sampling techniques have been proposed. In this paper, we focus on studying the performance of these techniques on Imbalanced Data of Arabic Sentiment. We then conduct a comparative evaluation using Support Vector Machine (SVM), Naive Bayes (NB), and Random Forest (RF) as classification algorithms.